Abstract:
This study presents an off-line signature verification system. The verification is based on geometrical and grid features through conducting two tests. The first one uses geometrical features using four border points in the direction of right and left, up and down in addition to two lengths.
The first connects the highest point with the lowest, and the second connects the left point with the right. The second test uses grid features taken from the core of a signature image.
The samples (population for this study) is drawn from 360 personal signature collected from 60 Arabic native speakers. The data that had obtained (along with one signature specimen) is cached in Oracle Database.
Practical experiments showed that the accuracy of the algorithm under False Rejection Rate (FRR) scale is less than 0.16 for the combined methods is geometrical and grid features.
